A note on the fourth cumulant of a finite mixture distribution

The paper shows that the fourth cumulant of a finite mixture distribution might be decomposed into the mean of the components’ fourth cumulants and the fourth cumulant of the components’ means, when the mixture’s components have the same second and third cumulants. Statistical applications include robustness properties of likelihood-based testing procedures and kurtosis-based projection methods. Practical relevance of theoretical results in the paper are illustrated with two well-known data sets.